Improve handling of large cascade plots

p3analysis/plot/backend/__init__.py

@@ -10,6 +10,9 @@
    <https://github.com/intel/p3-analysis-library/issues/new/choose>`_.
 """
 
+import itertools
+import string
+
 __all__ = [
     "Plot",
     "CascadePlot",
@@ -51,3 +54,24 @@ class NavChart(Plot):
 
     def __init__(self, backend):
         super().__init__(backend)
+
+
+def _get_platform_labels(platforms: list[str]) -> dict[str, str]:
+    """
+    Returns
+    -------
+    dict[str, str]:
+        A mapping from platform names to unique labels.
+    """
+    if len(platforms) <= len(string.ascii_uppercase):
+        labels = string.ascii_uppercase
+    elif len(platforms) <= len(string.ascii_uppercase) ** 2:
+        labels = []
+        for x, y in itertools.product(string.ascii_uppercase, repeat=2):
+            labels.append(f"{x}{y}")
+    else:
+        raise RuntimeError(
+            "The number of platforms supported by cascade plots is "
+            + f"currently limited to {len(string.ascii_uppercase)**2}.",
+        )
+    return dict(zip(platforms, labels))

p3analysis/plot/backend/matplotlib.py

@@ -5,8 +5,6 @@
 :py:mod:`matplotlib` backend.
 """
 
-import string
-
 import matplotlib
 import matplotlib.patches as mpatches
 import matplotlib.pyplot as plt
@@ -17,7 +15,7 @@
 import p3analysis.metrics
 from p3analysis._utils import _require_numeric
 from p3analysis.plot._common import ApplicationStyle, Legend, PlatformStyle
-from p3analysis.plot.backend import CascadePlot, NavChart
+from p3analysis.plot.backend import CascadePlot, NavChart, _get_platform_labels
 
 
 def _get_colors(applications, kwarg):
@@ -57,6 +55,10 @@ def create_artists(
     ):
         artist = []
 
+        # Make adjustments for large numbers of platforms.
+        if len(self.labels) > 26:
+            width *= 1.5
+
         # Draw a box using the platform's assigned color
         name = orig_handle.get_label()
         color = self.colors[name]
@@ -127,13 +129,16 @@ def __init__(
                 "filled_markers",
             )
 
-        # If the size is unset, default to 6 x 5
-        if not size:
-            size = (6, 5)
-
         platforms = df["platform"].unique()
         applications = df["application"].unique()
 
+        # If the size is unset, try to pick a sensible default.
+        if not size:
+            if len(platforms) <= 26:
+                size = (6, 5)
+            else:
+                size = (12, 10)
+
         # Create a 2x2 grid of subplots sharing axes
         fig = plt.figure(figsize=size)
         ratios = [6, len(applications) * 0.5]
@@ -168,12 +173,7 @@ def __init__(
         plat_colors = _get_colors(platforms, plat_style.colors)
 
         # Choose labels for each platform
-        if len(platforms) > len(string.ascii_uppercase):
-            raise RuntimeError(
-                "The number of platforms supported by cascade plots is "
-                + f"currently limited to {len(string.ascii_uppercase)}.",
-            )
-        plat_labels = dict(zip(platforms, string.ascii_uppercase))
+        plat_labels = _get_platform_labels(platforms)
 
         # Plot the efficiency cascade in the top-left (0, 0)
         app_handles = self.__efficiency_cascade(

p3analysis/plot/backend/pgfplots.py

@@ -5,8 +5,6 @@
 :py:mod:`pgfplots` backend.
 """
 
-import string
-
 import jinja2
 import matplotlib
 import matplotlib.pyplot as plt
@@ -16,7 +14,7 @@
 import p3analysis.metrics
 from p3analysis._utils import _require_numeric
 from p3analysis.plot._common import ApplicationStyle, Legend, PlatformStyle
-from p3analysis.plot.backend import CascadePlot, NavChart
+from p3analysis.plot.backend import CascadePlot, NavChart, _get_platform_labels
 
 # Define 19 default markers for LaTeX plots
 _pgfplots_markers = [
@@ -139,7 +137,7 @@ def __init__(self, df, eff_column, size=None, stream=None, **kwargs):
             ).strip("()")
 
         # Build a dictionary of platforms to labels
-        plat_labels = dict(zip(platforms, string.ascii_uppercase))
+        plat_labels = _get_platform_labels(platforms)
 
         # Choose colors for each platform and then convert the dictionary to
         # RGB colors using the platform labels
@@ -162,9 +160,6 @@ def __init__(self, df, eff_column, size=None, stream=None, **kwargs):
                 app
             ] = f"{app_to_tex_name[app]}, thick, solid, mark={mark}"
 
-        # Choose labels for each platform
-        plat_labels = dict(zip(platforms, string.ascii_uppercase))
-
         # Set the number of rows in the platform key (if set)
         # NOTE: This is different to matplotlib because PGF plots uses columns,
         #       and then transposes (so columns become rows)